Rapid and accurate detection of the biomass potassium (K) content in biomass is crucial for mitigating ash deposition and fouling issues in biomass fuel combustion processes. Laser-induced breakdown spectroscopy (LIBS) offers a promising approach for rapid analysis of biomass elemental. However, the accuracy of LIBS detection is susceptible to chemical matrix effects. Particularly in the case of biomass, characterized by its complex composition, the absence of pertinent studies on effect mechanisms impedes the enhancement of analytical accuracy. In this study, we investigated and compared two types of matrix effects related to chemical properties in biomass samples (analyte forms and matrix composition). Firstly, the inconsistency of K spectral response among different chemical properties samples and the performance of univariate models were analyzed. The results indicate that, compared to the chemical forms, differences in composition are the dominant factor of chemical matrix effects. Moreover, the compositional content of biomass samples is analyzed to correlate the matrix effect in LIBS measurements to a chemical property of the specimen. It is indicated that differences in volatile and ash content may lead to variations in the plasma excitation process, resulting in distinct spectra. Finally, a knowledge-based regression approach was employed to attenuate chemical matrix effects, the main influencing factors identified were analyzed as a priori knowledge for variable selection and inputting them into a partial least squares model. And 13 real solid biomass fuels were measured, resulting in root mean square error of prediction (RMSEP), R2, and average standard deviation (ASD) of 0.99, 0.050%, and 0.001%, respectively. This study investigated the influence of matrix effects related to biomass chemical properties on LIBS measurements, achieving rapid measurement of K. It promotes the application of LIBS in solid fuels and provides a methodological reference for LIBS analysis of complex matrix materials.
Read full abstract